nature methods Coding of facial expressions of pain in the laboratory mouse Dale J Langford 1, Andrea L Bailey 1, Mona Lisa Chanda 1, Sarah E Clarke 1, Tanya E Drummond 1, Stephanie Echols 2, Sarah Glick 1, Joelle Ingrao 1, Tammy Klassen-Ross 2, Michael L LaCroix-Fralish 1, Lynn Matsumiya 1, Robert E Sorge, Susana G Sotocinal 1, John M Tabaka 1, David Wong 2, Arn M J M van den Maagdenberg 3,4, Michel D Ferrari 4, Kenneth D Craig 2 & Jeffrey S Mogil 1 Supplementary figures and text: Supplementary Figure 1 Supplementary Figure 2 Supplementary Figure 3 Accuracy and reliability of the MGS. Specificity of the MGS to pain versus other states. Detection of acetaminophen analgesia by the MGS.
Coder Average % Observations Supplementary Figure 1 a 100 80 60 40 20 0 Novice -- Experienced -- --- Low Res. --- High Res. false alarms misses correct rejections hits b 2.0 1.5 1.0 0.5 ICC avg = 0.90 0.5 1.0 1.5 2.0 Overall Average Supplementary Fig. 1. Accuracy and reliability of the Mouse Grimace Scale. (a) Signal detection of novice and experienced coders on the abdominal constriction test, and the improvement in accuracy obtained using a high-definition (1920 x 1080) video camera (High Res.), using a selection of 64 randomized (pain and no pain) photographs. (b) Interrater reliability of the MGS scale. The mean MGS scores of each of six novice coders were compared to the average of all coders. ICC avg = average intraclass correlation coefficient (see Online Methods).
Supplementary Figure 2 A. Pain B. Illness - - C. Sleep n.s. - D. Grooming - Supplementary Figure 2. Mean MGS difference scores by action unit (Eyes: orbital tightening; Nose: nose bulge; Cheek: cheek bulge; Ears: ear position; Whiskers: whisker change; see main text for details) and overall average score for mice (A) injected with 0.9% acetic acid, (B) injected with 100 mg/kg lithium chloride (a dose effectively producing conditioned place aversion 1 ), (C) sleeping, and (D) actively exhibiting grooming behavior. Bars represent mean ± s.e.m. difference scores. Significantly different from zero by one-sample Student s t-test (P < 5, P < 1, P < 1). n.s., not scored. Reference 1. Tenk, C.M., Kavaliers, M., & Ossenkopp, K.-P. Eur. J. Pharmacol. 515, 117-127 (2005).
Withdrawal Threshold (g) MGS Score (0-2) Supplementary Figure 3 a 1.5 1.2 0.9 0.6 Baseline Pre-ACET Post-ACET 0.3 Vehicle ACET b 1.5 Baseline 1.2 Pre-ACET Post ACET 0.9 0.6 0.3 Vehicle ACET Supplementary Figure 3. Detection of the analgesic efficacy of 300 mg kg -1 acetaminophen (ACET) against zymosan inflammatory nociception by the MGS (a) but not von Frey withdrawal testing for mechanical allodynia (b). Separate groups of mice (n = 4-8/group) were tested at baseline (for details see Online Methods), then injected into the plantar hindpaw (20 l) with 5 mg ml -1 zymosan (100 g). From 4-4.5 h post-zymosan, mice were retested, and then injected with 300 mg kg -1 ACET or vehicle (polyethylene glycol; 10 ml kg -1 ). From 30 60 min post-acet, mice were tested yet again. Bars represent mean ± s.e.m. MGS scores (a) and withdrawal thresholds in g (b). As can be seen, ACET reduced MGS scores by approximately 50% compared to baseline, but was wholly ineffective against mechanical allodynia, which continued to increase in both vehicle and ACET-treated groups. Note that the 50% reversal of MGS scores is likely an underestimate, because vehicle group data in both experiments reveal the increasing intensity of the noxious stimulus from the pre-acet to post-acet time points. One-between (drug) repeated measures ANOVAs revealed a significant drug x repeated measure interaction for MGS (F 2,16 = 17.2, P < 1), but only a significant effect of repeated measure on von Frey withdrawal thresholds (F 2,20 = 28.0, P < 1). P < 5, P < 1, P <
1 compared to within-drug baseline, P < 1 compared to within-drug Pre-ACET (posthoc test for repeated measures; Systat v. 11).